We have a number of Web Logic instances running on same machine. For each Web Logic instance, we have an agent. The APM Agent on the same machine report a different value for CPU Utilization for the same core. Attach is an image to explain it better.
Release : 10.7.0
Component : APM Agents
We may be facing the limitation of the product (APM Agent). All APM Agents are running independent of each other. APM Agents that are monitoring Web Logic instances cannot run host-monitoring component of the agent to collect the CPU data at exact same instance (time). Thus, we see this discrepancy between the values obtained by different agents.
The nature of the CPU measurement fluctuates even from second to second. We cannot control the exact time when APM Agent polls for host monitoring. Thus, it is not possible to setup this on APM Agent side.
To handle this discrepancy between the values obtained by different agents on manager side, the following would be the most reliable and logical way to work around this.
Use management module calculator and determine the average value for values obtained by different agents for same CPU core. Then if you like to setup an alert, setup alert on this average value. Since CPU value can go high for few seconds, thus set the threshold value on 3/4 consecutive values when they stay above certain value. This can be setup under simple alert. This should help you setup CPU monitoring well.